erichiggins / gaek Goto Github PK

The auto-generated https://erichiggins.github.io/gaek/ site has the below link to a readthedocs page, which doesn't exist:

I don't think I can submit a PR to update this text since it looks like it's auto-generated from an older version of the gaek readme.

Benchmark against stock json module on Python 2.7:

Running json benchmarks...
Running ndb_json benchmarks...

Results
=======
loads (large obj)
-----------------
json            1.103s
ndb_json        2.200s

dumps (large obj)
-----------------
json            0.559s
ndb_json        0.553s

loads (small objs)
------------------
json            1.201s
ndb_json        2.760s

dumps (small objs)
------------------
json            0.913s
ndb_json        1.253s

The environment variables different when running Managed VMs, and the RPC stubs used by google.appengine.api.modules are not available. It would be beneficial for gaek to still function in this runtime environment as much as possible without throwing exceptions. Here's a dump of os.environ from a module running in Managed VMs with a custom image:

{
  'MEMCACHE_PORT_11211_TCP': 'tcp://172.17.0.3:11211',
  'SERVER_SOFTWARE': 'gunicorn/19.4.1',
  'MEMCACHE_PORT_11211_TCP_ADDR': '172.17.0.3', 
  'GAE_PARTITION': 's',
  'GAE_VM': 'true', 
  'MEMCACHE_NAME': '/gaeapp/memcache', 
  'HOME': '/root', 
  'GAE_MODULE_VERSION': 'vmv8', 
  'APPENGINE_LOADBALANCER_IP': '', 
  'MEMCACHE_PORT_11211_TCP_PORT': '11211', 
  'GAE_LONG_APP_ID': 'jt-calc', 
  'DEBIAN_FRONTEND': 'noninteractive', 
  'MEMCACHE_PORT': 'tcp://172.17.0.3:11211', 
  'MODULE_YAML_PATH': 'module.yaml', 
  'GAE_MINOR_VERSION': '389176423410002280', 
  'USE_MVM_AGENT': 'true', 
  'HTTP_X_APPENGINE_CURRENT_NAMESPACE': 'production', 
  'MEMCACHE_ENV_DEBIAN_FRONTEND': 'noninteractive', 
  'GAE_MODULE_NAME': 'mvm', 
  'GAE_APPENGINE_HOSTNAME': 'jt-calc.appspot.com', 
  'APPENGINE_LOADBALANCER': '', 
  'HOSTNAME': '90e692391545', 
  'GAE_AFFINITY': 'true', 
  'PWD': '/home/vmagent/app', 
  'GAE_MODULE_INSTANCE': '0', 
  'MEMCACHE_PORT_11211_TCP_PROTO': 'tcp',
# I injected this in the application init method based on GAE_MODULE_VERSION. 
  'CURRENT_VERSION_ID': 'vmv8',
# This was injected through my module.yaml file.
  'CURRENT_MODULE_ID': 'mvm', 
# These came from the Dockerfiles.
  'PORT': '8080', 
   'LANG': 'C.UTF-8', 
  'VIRTUAL_ENV': '/env', 
  'PKG_PATH': '/env/local/lib/python2.7/site-packages/', 
  'PATH': '/env/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin', 
  'SDK_VERSION': '1.9.30', 
  'SDK_URL': 'https://storage.googleapis.com/appengine-sdks/featured/google_appengine_1.9.30.zip', 
}

As noted in #20

Cloud NDB is a client library designed to replace App Engine NDB in apps that migrate from the App Engine Python 2.7 standard runtime to the Python 3 standard runtime.

As noted in #20 :

Based on an examination of the migration notes, it appears that there is at least one change which would break gaek, specifically the removal of google.appengine.api.namespace_manager which has no replacement.

The migration notes state:

There is no replacement for google.appengine.api.namespace_manager which is used to determine the default namespace when not passed in to Key()

This seems to primarily affect envrion.py which imports namespace_manager and makes its methods available.

Cloud NDB is a client library designed to replace App Engine NDB in apps that migrate from the App Engine Python 2.7 standard runtime to the Python 3 standard runtime.

Updating gaek to be cross compatible with both App Engine NDB and Cloud NDB would be extremely helpful to anyone who makes use of gaek currently, and wishes to move to Cloud NDB, while continuing to use gaek.

Based on an examination of the migration notes, it appears that there is at least one change which would break gaek, specifically the removal of google.appengine.api.namespace_manager which has no replacement.

Goal: Make gaek compatible with Python 3 (while maintaining 2.7 support)

As suggested by #20

Python versions supported by Google App Engine:

• 2.7 (Standard & Flexible)
• 3.6 (Flexible)
• 3.7 (Standard)

Docs for Porting Python 2 code to Python 3